Author: Sam Altman

Translation: Deep Tide TechFlow

Deep Tide Introduction: Someone threw a Molotov cocktail at Sam Altman's house at 3:45 AM. He rarely shares family photos publicly, hoping this might discourage the next person. This article is not just a denunciation of the attack; he also lays out his beliefs about AI for the first time: AI must be democratized, a few laboratories should not determine the future of humanity, and the temptation of "once you see AGI, you can't go back" has filled this field with Shakespearean drama.

This is a photo of my family. I love them more than anything.

I hope the image has power. Usually, we maintain considerable privacy, but in this case, I'm sharing a photo in hopes it might dissuade the next person from throwing a Molotov cocktail at our home, no matter how they see me.

The first person did this last night at 3:45 AM. Fortunately, it bounced off the house, and no one was injured.

Words also have power. A few days ago, there was an incendiary article about me. Yesterday, someone told me they thought the article came out at a time when people are extremely anxious about AI, making my situation more dangerous. I didn't think much of it then.

Now, I’m awake in the middle of the night, very angry, thinking about how I underestimated the power of words and narratives. This seems like a good time to talk about a few things.

First, my beliefs.

Striving for prosperity for everyone, empowering all individuals, and advancing science and technology are moral obligations for me.

AI will be the most powerful tool for extending human capabilities and potential. The demand for this tool is essentially limitless, and people will do incredible things with it. The world should have abundant AI, and we must figure out how to achieve that.

Not everything will go smoothly. Fears and anxieties about AI are reasonable; we are witnessing perhaps the largest societal change in a long time. We must ensure our safety, which is more than just aligning a model—we urgently need a societal response to new threats. This includes new policies to help navigate difficult economic transitions toward a better future.

AI must be democratized; power cannot be overly concentrated. Control of the future belongs to everyone and their institutions. AI needs to empower individuals, and we need to collectively determine our future and new rules. I don’t believe it’s correct for a few AI laboratories to make the most important decisions about the form of our future.

Adaptability is crucial. We are all learning new things at a very rapid pace; some of our beliefs will be right, some will be wrong, and at times we need to change our minds quickly as technology develops and society evolves. No one yet understands the impact of superintelligence, but it will be tremendous.

Second, some personal reflections.

When I look back at my work in the first decade at OpenAI, I can identify many things I am proud of and a slew of mistakes.

I’m thinking about the upcoming trial with Elon and recalling how staunchly I resisted agreeing to his desire for unilateral control over OpenAI. I am proud of that, proud of the narrow path we walked at that time that allowed OpenAI to continue existing, along with all subsequent achievements.

I am not proud of my avoidance of conflict, which brought great pain to me and OpenAI. I am not proud of how I mishandled the conflict with the former board, which caused significant chaos for the company. I made many other mistakes along the crazy trajectory of OpenAI; I am a flawed person, caught in an extraordinarily complex situation, trying to get a little better each year while always working for the mission. We knew from the beginning how great the risks of AI were, and that personal differences among well-meaning people I cared about would be greatly magnified. But experiencing those intense conflicts firsthand and often having to arbitrate them is another matter, and the costs have been severe. I apologize to those I have hurt, hoping I can learn faster.

I am also very aware that OpenAI is now a major platform, not a small startup, and that we need to operate in a more predictable manner now. The past few years have been extremely tense, chaotic, and high-pressure.

However, what I am most proud of is that we are fulfilling our mission, which seemed extremely unlikely when we started. Overcoming all obstacles, we figured out how to build very powerful AI, understood how to accumulate sufficient capital to build the infrastructure to deliver it, discovered how to establish product companies and businesses, and learned how to deliver services that are relatively safe and robust on a large scale, among other things. Many companies claim they want to change the world; we really did it.

Third, some thoughts about this industry.

I trace my personal lessons from the past few years, as well as my perspective on why there is such Shakespearean drama between the companies in our field, to this point: “Once you see AGI, you can’t unsee it.” It has a real "ring of power" dynamic that drives people to do crazy things. I’m not saying AGI itself is the ring, but rather the authoritarian philosophy of "being the one who controls AGI."

The only solution I can think of is to move toward widely sharing the technology with people and ensuring that no one owns the ring. The two obvious ways to achieve this are personal empowerment and ensuring that democratic systems remain in control.

It is essential that the democratic process remains stronger than companies. Laws and regulations will change, but we must operate within democratic processes, even if it will be messy and slower than we hope. We want to be a voice and a stakeholder, but not hold all the power.

Much of the criticism of our industry comes from genuine concerns about the extreme risks of this technology. This is entirely reasonable, and we welcome good-faith criticism and debate. I can empathize with anti-technology sentiments; clearly technology is not always good for everyone. But overall, I believe technological progress can make the future unbelievably good for your family and my family.

As we engage in that debate, we should de-escalate rhetoric and tactics, doing our best to have fewer explosions at fewer homes, whether in a metaphorical or literal sense.